stand/ficl 64-bit compatibility

[bdragon]

Summary

Currently, the only thing that prevents a functioning 64-bit ficl build is a few integer types that were intended to be fixed-width.

Changing them to C99 integer types allows building a functioning 64-bit ficl.

I tested this a while back locally the last time I was working on a Petitboot loader.

[imp]

stand/ficl/ficl.h

252

This won't work quite right for long long types though. Eg, it won't be all 1s

[bdragon]

stand/ficl/ficl.h

252

I believe that since FICL_INT is actually required to be equal to the size of void*, this is not an issue in practice.

[tsoome]

stand/ficl/ficl.h

252

if words are 64-bit and long long actually 128, then forth has double words for 128 bit

[bdragon]

stand/ficl/ficl.h

252

Yes, but that is a multiple-cell situation. The entire FICL codebase assumes that cells are either 32 bit or 64 bit and will not build if this is not the case.

[tsoome]

stand/ficl/ficl.h

252

Yes, exactly.

[bdragon]

stand/ficl/ficl.h

252

By the way, here are the overrides that these changes are in support of:

CFLAGS+=        -DFICL_INT=int64_t -DFICL_UNS=uint64_t -DBITS_PER_CELL=64 -DFICL_ALIGN=3

(for ficl running on powerpc64(le) in 64 bit mode)

Actually, reading this over again, I think the FICL_INT part is wrong too, it should have been FICL_UNS.

I guess I'll be adjusting this after all.

[imp]

stand/ficl/ficl.h

252

You need to cast the 0 to the proper type BEFORE ~ to get the right answer when FILC_INT is larger than long.

That's the point I'm trying to make. If there's casts needed for other reasons, you need to do those too.

[bdragon]

Addressed comments.

This version passes make tinderbox locally, and hardware test passed on G4 (memmap command works as expected)

[tsoome]

stand/powerpc/ofw/main.c

66

wouldn't it be more correct to use uintptr_t here?

[bdragon]

Address review comment